The total concentrations of valproic acid were higher in cord serum than in the serum of epileptic mothers given this drug (fetal/maternal total concentration ratios 1.7 +/- 0.5). The maternal free fractions of VPA correlated with the fetal accumulation of the drug. The fetal/maternal free fraction ratios (0.47 +/- 0.24) correlated inversely with the fetal/maternal total concentration ratios. The free concentrations of VPA in fetal blood were similar to those in maternal blood. These results obtained in vivo were confirmed by an in vitro study in which the drug had been added to drug-free serum samples. The free fractions (x100) of VPA in the maternal serum at birth (27.3 +/- 6.3) were significantly higher than in the serum of adult controls (8.0 +/- 2.4) and in cord serum (11.8 +/- 1.3). The pattern of VPA metabolite binding in the three groups was similar to that of VPA, although an unsaturated metabolite (2-en) was bound to a much higher degree than VPA (greater than 98%). The high total drug load in the fetus was partially displaced from binding sites during the first few postnatal days. The free fractions of the drug and metabolites in the neonates were almost twice as high as those in the fetus at birth. The decreased protein binding of VPA in the mothers at birth and in the neonates during the first postnatal week was related to increased free fatty acid levels. VPA concentrations in mother's milk were much lower than the free concentrations in plasma milk/plasma ratios 0.025 +/- 0.01). In neonates, half-lives for VPA were prolonged (43 +/- 14 hours). Our results indicate that increased free fatty acid concentrations in the maternal blood at the time of birth result in partial displacement of VPA from maternal binding sites, additional placental transfer, and thus fetal accumulation of the drug. The high drug load in the fetus is subsequently partially displaced after birth, resulting in increased free fractions and free concentrations of VPA in the neonate.